A Call for Founders: The AI Execution Layer of Supply Chain

The Forum Ventures Studio recently partnered with the Integrated Business & Engineering Program at Purdue University to explore new opportunities in one of our favorite spaces: supply chain. It’s a space we know well, as we’ve backed nearly 40 startups in the space, including Silq, Burq, and BetterSea.

Through in-depth research and conversations with operators, we identified nearly a hundred high-potential startup ideas grounded in real-world pain points.

Our research made one thing clear - we’re at an inflection point. Tech has finally caught up to the complexity of real-world operations, and a new generation of agentic AI will transform the supply chain industry from end to end. 

Specifically, we found one layer of the tech stack that remains completely underdeveloped: the execution layer. It’s the missing link between AI recommendations and real-world action. Most tools today can tell you what to do, very few can actually do it. The execution layer changes that.

What is the Execution Layer: It’s the middleware that lets AI agents move from insight to action. It enables them to:

• Interface with third-party systems (APIs, databases, browsers, terminals)
• Maintain state and execute sequences reliably
• Handle permissions, retries, rate limits, and fallback logic
• Operate across multi-step workflows with memory and context

This is the critical infrastructure that will turn AI from an assistant into an operator, capable of handling the messy, manual, and error-prone work that still dominates enterprise operations.

Here are our top 5 ideas from the research that we’re excited about and are looking for founders with deep domain expertise to co-build with, with full context, market and opportunity for each below:.

1. AI-Powered Certification and Origin Auditing
2. Autonomous HazMat Classification
3. Co-Pilot for Exception-Based Gaps
4. Procurement Agent for Custom Manufacturing
5. Agentic AI for Serial Number Automation

If you’re interested in building with in Supply Chain with Forum please apply through the founder-in-residence posting. Please highlight relevant experience and the area you’re most interested in.

AI-Powered Certification and Origin Auditing

Context
Enterprises in asset-heavy sectors like aerospace, defense, and industrial manufacturing face a persistent bottleneck in supplier compliance when it comes to validating certifications and country of origin claims. These workflows still rely on manual processes like extracting serial numbers from images, parsing scanned documents, and verifying supplier-reported information across outdated systems. As compliance pressure rises from regulations like Buy America and UFLPA, companies need faster, more accurate ways to validate sourcing and ensure supplier diversity metrics are met.

The Market
The intensifying regulatory scrutiny on global supply chains is already having tangible consequences. In the final quarter of 2024, U.S. Customs and Border Protection reported a 1,600% surge in detained shipments within the automotive and aerospace sectors compared to the same period in 2023. November alone saw 523 detentions, a dramatic increase from just 45 in August. In early 2025, CBP inspected 2,918 Chinese shipments for UFLPA compliance, 96% of which originated from the aerospace and automotive industries. Of those, more than 2,000 were denied entry.

The global market for supply chain compliance and traceability software is projected to surpass $9.5 billion by 2031, growing at a 14% CAGR. The category is growing fast, driven by stricter regulations and the high cost of non-compliance.

The Opportunity

The core challenge in supplier compliance, especially around certification and origin auditing, is not just knowing what to check, but turning collected data like certificates, part numbers, and origin documents into consistent, verifiable outcomes. Audits often reveal that origin documents are submitted but not properly validated, leaving importers exposed to penalties, delays, and disruptions.

To address this, an agentic execution layer could automate origin validation, check claims against trade rules, flag risks like non-originating inputs, and generate audit-ready records from the start, making the process proactive rather than reactive. Instead of scrambling through PDFs and emails during audits, enterprises gain continuous, automated compliance with a complete digital audit trail.

Why We’re Excited
As supplier diversity, compliance, and ethical sourcing become board-level priorities, enterprises will need intelligent systems to fill the gap between policy and operational enforcement. This is a whitespace where human labor and risk still dominate, and where automation can reduce costs while increasing auditability. 

Autonomous HazMat Classification

Context
Shipping hazardous materials like batteries, chemicals, aerosols, or industrial parts requires precise classification, documentation, and routing logic governed by complex global regulations. This causes logistics and Environmental, Health, and Safety teams to manually reference PDF guides, search regulatory databases, and complete Dangerous Goods Declarations by hand. One mistake can result in shipment rejection, six-figure fines, or safety violations.

The Market
In the U.S. alone, over 1 million hazardous materials shipments move daily, accounting for 12% of all freight tonnage. Globally, air cargo sees more than 1.25 million dangerous goods shipments each year, and that number is growing. Most existing solutions are still rules-based or manual, with no AI-powered system to handle classification and documentation from end to end.

The Opportunity
Build an agentic AI system that autonomously classifies hazardous materials, generates and submits compliance documentation, and initiates downstream execution steps in ERP, TMS, or customs platforms. This agent will read order data and SDSs, assign UN numbers and packaging instructions, flag violations, and push decisions into shipment workflows, all without human intervention. It will act not just as a decision-support tool, but as an autonomous compliance operator embedded in daily HazMat logistics

Why We’re Excited
This is a high-friction, high-frequency workflow that is begging for automation. The regulatory burden is only increasing, and the cost of mistakes is enormous. Yet no dominant solution exists to bridge the knowledge locked in PDFs and tribal expertise with action at the system level. We see this as a strong wedge into the market, with clear potential to expand into broader shipping documentation and compliance workflows. It’s a path to building the first verticalized, execution-capable AI agent for hazardous goods, transforming how they're classified, shipped, and tracked.

Co-Pilot for Exception-Based Gaps

Context
Despite decades of investment in enterprise software and automation tools, supply chains still fall apart when things don’t go as planned. Partial shipments, scheduling issues, pricing changes, or delivery delays usually trigger a chain of emails, phone calls, and manual updates across multiple systems, procurement tools, inventory management software, and transportation platforms. 

The Market
Global logistics and supply chain-related activities, including transportation, warehousing, and procurement, make up as much as 10–12% of global GDP, making it one of the largest sectors globally. Across such a massive, complex system, disruptions are bound to happen. Procurement automation software alone is projected to grow from $4.7B to $11.4B by 2030. And over 60% of procurement leaders say manual exception handling is one of their biggest efficiency blockers.

The Opportunity
Our current thesis is to build a co-pilot for exception-based supply chain workflows. This platform will facilitate intelligent communication, real-time negotiation, and autonomous system updates across procurement and logistics operations. It will allow AI agents to interpret inbound emails or calls, negotiate changes to orders or delivery plans, confirm updates with partners, and automatically reflect those updates in ERP, WMS, or TMS platforms. The system will also ensure transparency through audit logs and provide human-in-the-loop review for high-value or edge-case exceptions.

Why We’re Excited
Supply chain is one of the largest and most complex markets in the world, and inefficiencies run rampant across the sector. While traditional software has helped, agentic AI unlocks a new level of automation, especially for work that was previously too unstructured or manual to handle. Exception handling is a perfect starting point as it’s a persistent, costly bottleneck, and solving it can dramatically reduce human error and speed up delivery times. It’s also a highly repeatable problem that affects companies of all sizes, from SMBs to the Fortune 50.

Procurement Agent for Custom Manufacturing

Context
Custom manufacturing procurement today is a slow, manual, and highly technical process. Buyers must interpret complex design specs from CAD files, match them to qualified suppliers, and manage RFQs through to purchase orders, often via email, spreadsheets, and institutional knowledge. These tasks are time-consuming, prone to error, and heavily reliant on individual expertise/network. As the complexity and scale of custom component sourcing increase, the gap between manual procurement workflows and modern operational needs is widening.

The Market

The global custom manufacturing services market exceeds $250 billion, with over $40 billion in North America alone. Procurement labor represents roughly 3–5% of this expenditure. Today, more than 30,000 procurement teams handle specialized component sourcing, facing increasing pressure to accelerate workflows and reduce costs.

Despite advancements in digital quoting and manufacturability analysis, crucial steps remain stubbornly manual:

• Supplier Discovery: Typically consumes 40+ hours of manual research per supplier, often spanning weeks or months of emails and spreadsheets.

• Spec Interpretation & Supplier Matching: Procurement specialists spend multiple days manually reviewing complex CAD files to match designs to supplier capabilities, creating significant bottlenecks.
• Purchase Order Management: Manually processing a single purchase order costs organizations an average of $100 and can add days of delay per procurement cycle due to error-prone, email-driven approvals.

These manual processes collectively can costs larger enterprises millions in unnecessary operational expenses and lost productivity.

The Opportunity
Build an AI-native procurement agent designed specifically for the complexities of custom manufacturing. In industries where sourcing relies on interpreting technical specs, coordinating with specialized suppliers, and managing fragmented processes, workflows remain slow and manual. This agent would leverage execution-layer AI to interface directly with CAD software, supplier databases, communication channels, and ERP systems, automating the entire procurement cycle. This would compress days of manual work into minutes of intelligent automation.

Why We’re Excited

Procurement remains the biggest bottleneck to scale in custom manufacturing. As procurement teams face relentless pressure to accelerate sourcing, and deliver measurable cost savings, manual processes and fragmented tools continue to hold them back. An execution-enabled AI agent that fully automates procurement tasks could fundamentally transform these workflows, turning days of tedious labor into minutes of automated action. We see this as a repeatable, high-impact infrastructure play, where the company that cracks intelligent procurement automation would own a highly defensible layer within industrial supply chains.

Agentic AI for Serial Number Automation

‍Context
‍Traceability has become a foundational requirement for quality, compliance, and accountability across modern supply chains. From serialized components in maintenance workflows to repair and recall tracking, organizations need full visibility into how parts move, change, and are handled over time. Yet many teams still rely on manual steps, including visually reading serial numbers, capturing images, and entering data into disconnected systems like SAP. This creates slow, error-prone processes that delay audits, hinder repairs, and introduce compliance risk, especially when data gaps can trigger regulatory penalties, safety incidents, or operational downtime.

‍The Market
‍Regulatory pressures and operational complexity are driving demand for automated part traceability across aerospace, defense, healthcare, and automotive sectors. Several recent enforcement actions highlight this growing need:

• Aviation (FAA/EASA): Over 90 aircraft grounded in 2023 due to counterfeit engine components lacking verifiable serial numbers, violating FAA and EASA traceability mandates.
• Automotive (NHTSA): Ford fined a record $165 million in 2024 by NHTSA for insufficient VIN-based traceability during recalls.
• Healthcare (FDA UDI): FDA intensified enforcement of Unique Device Identification (UDI) regulations in 2023–2024, citing multiple medical device manufacturers for incomplete serial number records.

Driven by these regulatory actions, stringent compliance requirements, and advancements in serialization technologies, the global product traceability software market is projected to reach $28.2 billion by 2026, growing at a CAGR of 8.8%

The Opportunity
Our current thesis is to build an agentic AI platform that automates the end-to-end process of reading serial numbers from images, validating them against enterprise databases, and taking action directly in systems like SAP. This platform will transform static image inputs into fully autonomous workflows, generating return purchase orders, updating asset registries, or logging maintenance events in real time. Designed for high-stakes environments, it will include a human-in-the-loop feature for edge cases while enabling fully autonomous execution in routine workflows.

Why We’re Excited
As enterprises adopt more advanced AI across workflows, this opportunity represents a foundational infrastructure play, particularly in industries like aerospace, healthcare, and automotive, where traceability is mission-critical. These sectors face intense regulatory pressure, yet still rely heavily on manual processes. This capability will be a necessary layer of infrastructure for managing traceability across the entire supply chain.

Final Thoughts

We're seeking exceptional founders to tackle these five concepts head-on. With the support of a dedicated team and $250K in initial funding, you’ll have everything you need to transform these concepts into thriving businesses.

If you’re interested in building with in Supply Chain with Forum please apply through the founder-in-residence posting. Please highlight relevant experience and the area you’re most interested in.

While we’re actively building out these ideas, Forum Ventures is always open to hearing from founders with bold, original ideas. If you're working on an early-stage B2B SaaS company, even at the idea stage, we’d love to hear your pitch through the Forum Studio Founder posting.